Start with the lesson you can demonstrate using material the client has approved for publication. An account of how a team solved a problem can be useful even when the client name, revenue and internal reports have to stay private. The published claim should be as specific as the approved evidence allows. If the evidence only supports a description of the method, publish the method and explain its limits.
Treat the approved public version as its own deliverable. Decide what the reader needs to learn, what the team can substantiate internally and what the client permits the reader to see. Those are separate decisions. A fact can be accurate, relevant and still inappropriate for public release. Plan the article around the intersection before writing a headline.
Disclosure. AI Visibility is published by Lunasol, which offers GEO and AEO services. The worked case below is fictional. The public examples are separate from that case. About AI Visibility
Define a lesson that survives the confidentiality limits
Write one sentence about what the case would help a reader decide. For example, a service manager might need to choose whether to clarify eligibility on a service page before changing the enquiry form. That lesson can be supported by an approved account of the questions customers asked and the steps the team took. A claim that the change increased qualified enquiries would need its own evidence and permission.
Make a list of the facts necessary to tell the story. Separate the starting problem, the intervention, the observation and the interpretation. If the client permits the problem and intervention but withholds all results, a methods article can still explain how to make the change and check it. State where the example ends. A headline that implies a proven outcome can give the wrong impression even if the body carefully avoids figures.
Review adjacent published material before commissioning. Lunasol publishes a named case study for DP Business Solutions that presents client-specific figures and a defined period. That is one public format. A confidential case needs its own evidence and release decisions. The visible named case also illustrates why a headline, client description and timeframe can carry identifying information even before the reader opens the rest of the page. Lunasol's published DP Business Solutions case study
Review permission at the level of each item
Keep an internal release record. For each fact, table, quote or image, name the source and the proposed public use. Record whether the client has approved the exact wording, approved a narrower version or held it back. Include the person responsible for the decision and the approved edition. Use the client's established review route and bring uncertain material to the person responsible for contractual or privacy review.
| Item | Internal evidence | Possible public treatment after approval |
|---|---|---|
| Problem | Support notes or research record | Describe the recurring issue in broader terms |
| Intervention | Change log and approved version | Explain the steps the team actually took |
| Outcome | Verified measurement and method | Release an approved statement at the supported level |
| Testimonial | Original wording and speaker record | Use the exact approved wording and attribution |
| Image | Original file, rights and context | Publish an approved asset or an illustration labelled as such |
Scope the approval to the formats in which people will encounter the story. A client may agree to a page with a neutral title while preferring a different decision about a social post, downloadable slide or sales deck. Record the channels that were reviewed. If an editor later changes the headline or introduces a new crop, send the changed item through the relevant review before release.
A public Instagram post can illustrate the distinction between visible context and permission to connect it to other facts. Dubai Fitness Challenge's previously verified 2 November 2026 reel looks back at Dubai Ride. A writer can cite the organiser's public recap when discussing that public event. Linking a confidential project to the same date, route, distinctive image or caption could expose a relationship that the recap itself never disclosed. The reel is an independent illustration and has no connection to the fictional case below. Official Dubai Fitness Challenge reel
Check what a reader could identify from the remaining details
Removing a name is one step. Review the combination of industry, location, dates, employee role, product description, unusual events and numerical changes. A supplier, competitor or former employee may recognise a pattern that means little to a general reader. Check whether the headline and description are more identifying together than either would be on its own.
The UK's Information Commissioner's Office discusses singling out and linkability when assessing personal information, and describes testing what a motivated person could infer using other material. That suggests a practical editorial exercise for a confidential case. Ask a reviewer to look at the candidate public details together and consider which ones could be connected to public announcements or prior articles. Client confidentiality and personal data questions need to be reviewed under the rules that apply to the project. ICO guidance on assessing anonymisation
Inspect the actual files as well as the prose. A cropped screenshot may show an account name in a tab or tooltip. The file name, image description, download properties or text behind a visual overlay can expose material the article omits. Review the public export separately and make sure any illustrations are created from approved content. A diagram recreated from scratch should be labelled if readers could mistake it for a captured client screen.
Keep the evidence and its transformation traceable
For each proposed result, record the original measure, period and unit in the restricted source record. Then record what the public version leaves out or changes. When using an approved rounded figure or a range, check whether it can be combined with another published figure to infer a restricted value. A percentage and a known count can reveal information that either figure alone would conceal.
A broader label should preserve the meaning of the original evidence. A change observed in a subset of enquiries cannot become a statement about the whole business simply because the subset name is sensitive. If the approved context cannot explain what the observation represents, use a narrower claim or omit the result. Keep the source, calculation and permission linked in the internal record so another authorised reviewer can reproduce the decision.
Qualitative evidence deserves the same treatment. An approved interview can support what the interviewee observed. A quotation needs the speaker's permitted wording and attribution. If several people reported similar issues, explain how those accounts were gathered at the level the client approved. Avoid presenting a reconstructed exchange as a verbatim conversation. A clearly labelled sample can show the reader how an interaction might work while keeping it separate from what actually happened.
Work through a disclosure decision
Imagine a fictional appointment-based business whose team wants to show how it clarified which enquiries needed an assessment. Its internal record contains old booking screens, staff notes and enquiry data. For this example, the permitted public account covers the problem and the revised instructions. The business name, screenshots, dates, counts and outcome figures are outside the approved material.
The team can explain how the page began to distinguish requests that could proceed directly from requests that needed a preliminary check. It can describe where the instructions appeared and what the staff member checked next. A reader can apply that method to their own service. The team should write the result as a description of the implemented change. A statement that the number of repeat enquiries fell would require approved result evidence.
Suppose the client later permits one broader observation, such as a recurring question appearing less often in reviewed support records. The writer would still need to check the scope of the records and whether the wording accurately reflects them. If the period and collection conditions are too restricted to make the observation interpretable, keep the original methods article. Save the revised permission decision in the release record regardless of whether the team uses it.
Give the example a title that matches the released evidence. “How a service team separated direct bookings from assessment requests” describes the intervention. A headline promising fewer enquiries would require a different evidential basis. Apply the same review to the introduction, caption and search description. Those short pieces can imply a result even if the full article presents a careful account.
Watch for clues introduced by the explanation of what was withheld. A sentence saying that the client operates the only site of its kind in a particular district can undo a carefully general description. Explain the relevant limitation in ordinary terms, such as “The example describes the booking instructions. Enquiry data remains private.” A reader can then judge the lesson without being given hints about the underlying organisation.
Give designers and distribution editors the same release record. A visual brief may ask for a recognisable building or for a mock dashboard with sample figures. Check those assets against the approved claims and make the illustrative status clear. The article, its previews and any sales material should present the same evidence boundary.
Verify the exact public package before release
Confidential case publication record
- Reader decision and supported lesson
- Internal evidence owner and approved version
- Public facts, assets and wording cleared for each channel
- Removed details and the reason for withholding them
- Remaining combinations that could identify a person or business
- Headline, captions, metadata and downloadable assets reviewed
- Approval owner, date and process for changes
Use separate reviews for accuracy and disclosure. Have an authorised reviewer trace the public claims back to the internal record. Have another reviewer inspect the article as a reader would see it, including images, file names, links and extracts. Ask each person to say what they think the case proves. If their answers differ from the claim the team intended to make, adjust the evidence or wording and review the new version.
Check related material that a reader can find from the page. A portfolio entry, earlier event announcement or staff post can supply context the case intentionally leaves out. Review how the links, dates and descriptions work together. If the connections create a problem that the team cannot resolve, publish a general guide with its own evidence or a labelled hypothetical example. That gives readers a useful procedure while keeping the client work private.
Finally, record exactly what went live and when it was checked. Keep a way for the owner to request an amendment if facts, permissions or circumstances change. An editor taking over should be able to see which part of the article is supported, who approved the released material and which claims would require a new review. That record makes future changes easier to assess without asking the client to repeat the original approval process.



